The Enhanced Color Image Based On The Total Variation (iv)

Image enhancement is a classical subject of image processing. As a significant pre-process, its related researches have been thoroughly implemented. Nowadays, applications of color images become more extensive. Therefore, researches about the enhancement of color image become more and more important.This thesis followed the basic idea of principal component enhancement to carry out the researches. In the first place, the study was to do the transformation of color image to gray image. Then, doing contrast enhancement of gray image was put on the procedure. Finally, the issue turned to be the reconstruction of color image from the enhanced gray image. The outcome of color image enhancement lies on the transformation of color image to gray image and the enhancement of gray image. Unfortunately, traditional and existed methods of color image enhancement have the problem of shape distortion or edge blurring.This thesis thoroughly researched the transformation of color image to gray image. Combined the concept of level-lines of vector-valued images and the theory of total variation, an improved method based on total variation image restoration model was presented. This new method achieved a better unity of shape-preserving and edge-protecting by unifying these two aspects in one model. It overcame the disadvantages of traditional and existed methods. Furthermore, this thesis compared and researched technologies of gray images' contrast enhancement. It utilized the shape-preserving local histogram equalization to do contrast enhancement of the principal component image of the original color image. It used the log enhancement to pre-process the gray image and pre-detected the acreages of connected components to determine the thresholds of operations. Finally, it implemented the iterations by multi-level hierarchical approach. The contrast of whole gray image was improved by a shape-preserving way. Besides, contrast of small objects in local and low gray areas was improved too. Ultimately, on the basis of above-mentioned two studies of this thesis, a new scheme of color image enhancement based on total variation was presented. Experimental results of both artificial and natural color images showed that the method advanced in this thesis was effective for color image enhancement and produced shape-preserved enhanced color images with edge sharpness much better than that of the existed method.